Thanks to a reader’s recommendation, I found my way to a paper about the concept of loudness normalization written by Florian Camerer. It’s from the later part of 2010 but provides a great deal of information about the “loudness normalization revolution!” You can get the article here. It’s called, “On the way to Loudness nirvana – audio leveling with EBU R 128”.
The EBU R 128 recommendation “establishes a predictable and well-defined method to measure the loudness level for news, sports, advertisements, drama, music, promotions, films etc., throughout the broadcast chain, and thereby helps professionals to create robust specifications for ingest, production, playout and distribution to a multitude of platforms.” It was developed and introduced because consumers of audio were bugged by the constant and sudden shifts in loudness between programs. We’ve all experienced it. You’re listening to a playlist on your portable player or broadcast device and suddenly there is a huge increase in overall volume as you move from the quiet tail out of a music ballad to a highly compressed and loudness maximized commercial. The reality is that playlists of music on Spotify, iTunes, Pandora, and YouTube have all suffered the same problem…until “loudness normalization”.
Well the European Broadcasting Union (the EBU) decided to do something about it and the result is the R 128 recommendation. It’s important to understand that this applies to broadcasts and doesn’t actually apply to individual music tracks or even music playlists. The distribution channels mentioned above are the ones that are implementing “loudness normalization” to smooth out their playlists. This idea is what has Bob Katz and Ian Shepherd so excited that they’ve written articles about the end of the loudness wars. I agree that it’s a step in the right direction but I’m not seeing it as a major step forward…at least not yet.
Let’s consider what the author of the paper calls, “one of the most fundamental changes in the history of audio in broadcasting: the change of the leveling paradigm from peak normalization to loudness normalization.
First, let’s consider what loudness is. “Loudness refers to the perceived strength of a piece of audio (music, speech, sound effects etc.). The loudness depends on the level, frequency, content and the duration of the audio, amongst other things.” One of the major roles of a mastering engineer is to make sure that all of tracks on an album are the same loudness…the listener should be able to jump from any moment in the program to any other moment in the program and not have to make any adjustments to the playback volume on their AVR. To achieve this goal, a mastering engineer has a number of tools and techniques at his or her disposal. You can simply turn up any tune that is perceived as too quiet and/or lower the loudness for any tunes that are perceived as too loud.
Over this past weekend, I mastered a wonderful new recording of Paul Horner singing 11 of his best tunes (he was a major songwriting collaborator of Peggy Lee) while accompanying himself on solo piano (I wrote about the session some months ago…read here). We recorded the project in real HD-Audio, but I did the mastering for the CD version. The first tune is a solo piano selection and contains a number of very loud moments. However, the rest of the tracks include Paul’s singing and are therefore lower in overall loudness than the instrumental opener. What was I supposed to do? I could lower the amplitude of the opening number but then the relative balance between the rest of the tracks would sound odd. In the end, I compressed the dynamic range of the first tune and turned up the relative volume of the vocal tunes. It worked wonderfully…as long as you experience the album in sequence.
But what happens when an individual tune (with plenty of the original dynamics left in tact) is isolated from the album and played individually? It’s going to sound wimpy against the rest of the songs in a playlist. Imagine hearing a Tony Bennett track just prior to one of Paul’s (he wrote a number of tunes that Tony sang BTW), the perceived loudness between the two would be objectionable.
The EBU R 128 recommendations takes care of the problem by “normalizing the loudness” of the two tunes resulting in them sound very similar in overall amplitude. Problem solved, right? No sadly…it’s not.
To be continued.